Google Invests $40B Into Anthropic, GPT 5.5 Drops, and Google Cloud Dominates | EP #252

May 4, 2026

AI Summary

5 min read

The Moonshots podcast recaps a whirlwind week in AI, highlighting new model releases like OpenAI's GPT-5.5 and China's Kimi K2.6, Google's $40 billion commitment to Anthropic, expansions in Google Cloud compute, and breakthroughs in health tech alongside robotics and government adoption. Hosts Peter Diamandis, Alex Wissgott-Gruber, Dave Blei, and Salim Ishmael emphasize the accelerating US-China model race, compute bottlenecks, and practical implications for enterprises and consumers.

Model Releases and Capabilities Surge

OpenAI launched GPT-5.5 seven weeks after GPT-5.4, touting it as a "new class of intelligence" with native omni-modality (text, audio, video, images in a unified architecture). Key gains include a 37-point jump in long-context reasoning over a million-token window, 40% token efficiency improvement, and 60% fewer hallucinations. It excels on Terminal Bench for agentic terminal operations, aiding tools like Codex, and Frontier Math Tier 4, with ~1-2% monthly gains signaling "math is cooked" for professional research problems within years. API pricing doubled to $5/million input and $30/million output tokens.

Continue reading the full summary in the app — free to try.

Read Full Summary →

Free • No credit card required

Listen to Audio Summary Open in App

What you'll learn

1 (00:00) **Google's $40B Anthropic Investment** - Google commits $40B ($10B now, $30B conditional) at $350B valuation, plus 5GW TPUs over 5 years
2 (00:17) **Google Cloud TPU v8 Dominance** - Unveils TPU 8T (training) and 8i (inference), 3x faster training, 80% better perf/dollar
3 (00:32) **GPT-5.5 Release** - OpenAI drops GPT-5.5 seven weeks after 5.4, native omni-modality, 37pt long-context gain, 60% less hallucination
4 (03:26) **AI Model Release Frenzy** - 15 major models in 8 weeks (Kimi K2.6, GPT-5.5, DeepSeek V4), US-China duopoly
5 (14:02) **Moonshot AI's Kimi K2.6** - 1T-param open-weight MoE model, 32B active params, multimodal, 30x cheaper than closed models
6 (21:38) **Mixture of Experts (MoE) Explained** - Kimi/DeepSeek use layer-by-layer MoE sparsity for efficiency, basis of frontier models
7 (31:10) **Compute Bottlenecks: TSMC/Energy** - All hyperscaler deals bottlenecked at TSMC fab capacity; energy/permitting key now

+ Full timestamped outline available in the app

Show Notes

In this episode, the mates explore the rapid advancements in AI, the global AI race, new models like Kimi K 2.6 and GPT 5.5, and the implications for privacy, security, and the future of technology. The mates discuss the future of work, AI regulation, and transformative biomedical breakthroughs.

Get access to metatrends 10+ years before anyone else - https://qr.diamandis.com/metatrends

Peter H. Diamandis, MD, is the Founder of XPRIZE, Singularity University, ZeroG, and A360

Salim Ismail is the founder of OpenExO

Dave Blundin is the founder & GP of Link Ventures

Dr. Alexander Wissner-Gross is a computer scientist and founder of Reified

–

My companies:

Apply to Dave's and my new fund:https://qr.diamandis.com/linkventureslanding

Go to Blitzy to book a free demo and start building today: https://qr.diamandis.com/blitzy

Your body is incredibly good at hiding disease. Schedule a call with Fountain Life to add healthy decades to your life, and to learn more about their Memberships: https://www.fountainlife.com/peter

Connect with Peter:

Connect with Dave:

Connect with Salim:

Join Salim's Workshop to build your ExO

Connect with Alex